Deep learning algorithms have become powerful means of addressing a very large class of problems, especially within the field of image processing. Due to a confluence of improvements in computational power, training algorithms and our ability to collect and process large quantities of data, it has become possible to automate tasks which were considered extremely challenging even a decade ago. These models are making a large impact on the way medical images are interpreted, leading to better patient diagnosis and improved therapy selection.
In this talk, I will summarize key examples of clinical applications where deep learning based algorithms are already improving reading and quantification of medical images. In the latter half, I will focus on recent research applying these models to capture the physics of blood flow in patient-specific anatomies. By utilizing the anatomic information of the geometry of a patient’s arterial network, it is possible to analyze the hemodynamics using well-known computational (CFD) techniques. However, such models are very demanding of computational resources, as well as require sufficient engineering expertise for getting reliable results. I will highlight recent research from our team showing how AI models could offer fast, automated, patient-specific hemodynamic analysis, reducing the need for needless invasive testing.
Sai Rapaka is a senior scientist working on AI algorithms at the Siemens Healthcare Technology Center in Princeton, NJ. His research covers two main areas: automatic quantification of medical imaging for cardiovascular disease characterization and patient-specific functional modeling. His work has resulted in multiple U.S. and global patents as well as publications. He received his Ph.D in mechanical engineering from Johns Hopkins University and was a postdoctoral research associate at Los Alamos National Laboratory prior to joining Siemens.
For opportunity to talk with the speaker please contact Kevin Connington at [email protected]